[TMA] Enable unswizzled tma layouts #6238
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
peterbell10
commented
Mar 18, 2025
| swizzlingByteWidth = 32; | ||
| } else { | ||
| llvm_unreachable("unsupported shared memory layout for MMAv3"); | ||
| llvm_unreachable("unsupported NVMMA layout (MMAv3 or TMA)"); |
Contributor
Author
There was a problem hiding this comment.
Not directly related, but seeing this MMAv3 message was a bit confusing on blackwell.
Contributor
There was a problem hiding this comment.
Yeah, it was dropped before the shared memory encoding refactoring
peterbell10
commented
Mar 22, 2025
| auto rank = tensorType.getRank(); | ||
| if (rank == 1 || contigDimSizeInBytes < 32 || shape[rank - 2] < 8) { | ||
| return ttg::SwizzledSharedEncodingAttr::get(ctx, 1, 1, 1, order, ctaLayout); | ||
| } |
Contributor
Author
There was a problem hiding this comment.
Okay, now defaulting to unswizzled encodings if there are fewer than 8 rows of data.
| // CHECK-DAG: #[[NVMMA_32:.*]] = #ttg.nvmma_shared<{swizzlingByteWidth = 32, transposed = false, elementBitWidth = 8}> | ||
| tt.func public @tma_scatter(%arg0: !tt.ptr<i8> {tt.divisibility = 16 : i32}, %arg1: i32 {tt.divisibility = 16 : i32}, %arg2: i32 {tt.divisibility = 16 : i32}, %arg3: tensor<32xi32, #blocked>, %arg4: tensor<32x32xi8, #blocked1>) { | ||
| // CHECK: tt.make_tensor_descriptor {{.*}} : <i8>, <tensor<1x32xi8, #[[NVMMA_32]]>> | ||
| // CHECK: tt.descriptor_scatter {{.*}} : !tt.tensordesc<tensor<1x32xi8, #[[NVMMA_32]]>>, {{.*}} |
Contributor
Author
There was a problem hiding this comment.
This test ensures that gather/scatter still use swizzling despite the descriptor's block shape only having 1 row.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.